There is a need for more powerful sources of extreme ultraviolet (EUV) light at 13.5 nm in order to increase the throughput of semiconductor patterning via the process of EUV Lithography. Many different source designs have been proposed and tested (see historical summary for background [1]) including the highly efficient (up to 30%) direct discharge (DPP) lithium approach [2,3,4,5,6,7] and also laser-plasma (LPP) irradiation of tin—containing [8] or pure tin droplets [9,10,11]. Laser irradiation of tin droplets has been the subject of intensive recent development [12,13], particularly in the pre-pulse variant [11], which has a demonstrated efficiency of 4% and a theoretical efficiency of up to 6%.
In both lithium DPP and tin LPP approaches it is necessary to keep metal atoms from condensing on the collection mirror that faces the EUV-emitting plasma. Also, in the tin LPP approach, but not with lithium DPP, there are fast ions ranging up to 5 keV that have to be stopped otherwise the collection mirror suffers sputter erosion. The design of a successful EUV source based on a metal vapor must strictly protect against deposition on the collector of even 1 nm of metal in days and weeks of operation, and this factor provides the most critical constraint on all of the physics that can occur in a high power source. In the case of lithium, extremely thorough metal vapor containment is provided via a buffer gas heat pipe [2]. However, the heat pipe containment technology cannot be extended to tin sources because the heat pipe temperature would have to be 1300° C. to provide the equivalent tin vapor pressure versus 750° C. for lithium. This vastly higher working temperature renders the heat pipe approach essentially unworkable for tin whereas it is very practicable for lithium.
Harilal et al. [14,15] have performed a series of studies on the use with a tin LPP source of either a magnetic field, a buffer gas, or a combination of these to slow down fast ions and protect the collection optic. Many magnetic field configurations have been discussed [16-29], with and without a buffer gas, to trap and exhaust tin ions. Methods have been proposed [30,31] to further ionize tin atoms so that they may be controlled by an applied magnetic field. The symmetrical magnetic mirror trap [18] has a long axial exhaust path for tin ions and if this path has a shallow gradient of magnetic field, can suffer from a build-up of plasma density as successive tin droplets are irradiated. Two things begin to go wrong: 1) there is an EUV absorption cross section of 2×10−17 cm2 for tin atoms that causes increasing EUV absorption loss as the plasma density builds, and 2) the mirror magnetic trap is unstable [14] to lateral plasma loss, which can expose the collection optic to tin atoms. Refinements of the mirror trap have been described [20,23] in which an asymmetry is introduced so that plasma flow is toward a weaker magnetic field at one end of the mirror configuration. This also can be combined with an electric field [20] to aid plasma extraction at the end with lower magnetic field. However, only a relatively constricted path is available for plasma exhaust toward one end of such a trap configuration, implying a limited heat removal capacity. Other magnetic configurations [27,29] have been designed to protect the collection optic, but these rely on gas cooling, and do not provide a specific path for plasma flow toward a large area plasma beam dump. Accordingly, the power scaling of such configurations is limited due to lack of heat removal.
Buffer gases have been discussed [15,32,33] to reduce ion energy and protect the collection optic. One of the main buffer gases used has been hydrogen [13,33] but as plasma power increases there is an increasing fraction of molecular hydrogen dissociation that can lead to vacuum pumping and handling problems of reactive hydrogen radicals. Coolant gases with more favorable properties, in that they do not react chemically, are argon and helium. These gases have higher EUV absorption than hydrogen [15], so they may only be used at lower density. However, argon has substantial stopping power for fast tin ions [15], and is particularly effective when a magnetic field is combined with a gas buffer to lengthen the path of tin ions via curvature in the field.